Cellular and PCS operators have eked out gains over the years-to the point where the gains curve has flexed asymptotically, giving minimal return for maximum effort. As a result, many operators have cast their nets wider, hauling in such advanced technologies as smart antennas and multiuser detection (MUD) to implement in their basestations. As chip geometries fall to 0.13 micron, the power of the processors needed to realize these two promising, compute-intensive technologies is now becoming available and the cost is dropping. Combined with streamlined, efficient algorithms, smart antennas and MUD are on the cusp of full realization.
Smart antennas come in various categories, ranging from basic switched-beam to fully adaptive antennas. However, all use some level of processing to focus the antenna's beam to maximize signal reception-and in more advanced cases, null out interferers. For its part, MUD, which is used only in code-division multiple-access (CDMA) systems to date, employs the principle of joint detection to detect the characteristics of co-users of a channel and then uses a subtractive process to cancel them out, leaving the target signal.
Though each is capable of providing a 3x or more capacity/performance improvement per basestation, thanks to the pioneering efforts of companies such as Metawave Communications Corp. and Arraycomm Inc., the smart antennas is the first technology of the two to have reached any level of commercial success. With a frequency-division-duplex focus, which makes its products amenable to Global System for Mobile Communications, CDMA and wideband CDMA networks, Metawave has sold more than $150 million worth of smart-antenna equipment to operators such as Sprint and Verizon.
Smart antennas are getting embedded within the basestation, as illustrated by the deal Metawave just signed with Samsung.
All-in-one basestation?
From a bird's-eye view then, the obvious trend is the compaction into the signal-processing domain of such highly compute-intensive technologies as smart antennas, MUD, software-defined radio and DSP-based digital predistortion for multicarrier power amplifiers. The goal is to ultimately develop a low-cost, low-power, highly flexible programmable basestation that can adapt dynamically to the operator's needs in terms of capacity, coverage and service quality. While an all-in-one solution is still many years off, the foundation is being laid right now for such a future, thanks to the processing and scalable-architecture options currently being offered.
Processing power alone is not the gating factor for smart antennas or MUD. As many are now finding out, the key turns on how that processing power is distributed among DSPs, ASICs, FPGAs and memory. As a result, Mercury Computer Systems (Chelmsford, Mass.) found Intrinsity's use of a 4-Gbyte/second RapidIO interface particularly attractive, and has since declared itself a partner with Intrinsity. For its part, Intrinsity (Austin, Texas) liked Mercury's MUD algorithms.
Recently Ericsson has pushed for MUD standardization, meaning that its own MUD intellectual property could well form the basis of that standard-if it jells. But such an outcome does not bode well for Mercury or others with proprietary MUD IP.
While Intrinsity looks for a toehold in the market, Texas Instruments Inc., whose C6416 is designed into eight out of 10 upcoming 3G basestations, is continuing to promote the advantages of its DSP/ASIC architecture. However, TI too is cognizant of the importance of moving beyond its 4-Gbit/s external-memory-interface I/O port, and might well be considering the flexible, scalable switching fabric of RapidIO.
Analog Devices Inc., with its TigerSharc DSP, is uncertain whether 3G systems will incorporate smart antennas or MUD out of the chute, citing low initial capacity concerns for operators that will be more concerned with getting the systems to work. However, unlike second-generation and 2.5G systems, which depend on Metawave and Arraycomm-like add-ons, 3G systems have hooks for those technologies built in, allowing their incorporation into the baseband-processing section of the basestation.
Once embedded within the system, the potential of both technologies can be maximized. While MUD is strictly an uplink enhancement, smart antennas enhance the uplink and downlink and will soon be found working with multicarrier power amps to take advantage of MUD's channel-estimation features and sector-use knowledge to lower power amp consumption and usage. Regardless of the processing architecture, the move to an all-software basestation design will have then begun to take hold.
